A corpus must be annotated in order to be queried by a computer. The annotation task, which is extremely time-consuming and rigorous, involves each sign being given a single label. This label or ID-gloss means that all the uses of the same sign, or more precisely of the same lemma (a lemma gathers all the morphological variations of a sign, such as the infinitive in French that gathers, in a dictionary, all the conjugated forms of a verb), can be found in the corpus.

The ID-gloss of a sign must not be mistaken for the French translation of the sign. The gloss can be a number or an arbitrary symbol. However, French words (or combinations of words) have been used as glosses for ease of annotation. They are written in capital letters.

So far, more than 10 hours of video have been annotated sign by sign. These annotations are available to registered users. When a video is annotated, the name of the task contains the label “annotated” (list view) or the label “(A)” (button view).

Partly-lexicalised signs (i.e. iconic signs whose meaning depends on the context) are provisionally annotated as “DS” for depicting sign*. In the near future, they will be completed with a description of their meaning in context.

The fully-lexical signs identified in the 10 hours of annotated video have been gathered in a lexical database (Lex-LSFB) that is also available on the website. This lexical database is not a dictionary. It is a list containing all the signs identified in the annotated videos, with their gloss and some possible French meanings. This lexical database will be regularly enlarged as the corpus annotation continues.

These sign glosses and the French keywords linked to the lexical database can be used to search for videos (e.g. which videos contain the sign EXACT? Which videos contain signs that mean “precise” or “correct” in French?).

* See the annotation conventions of Australian Sign Language (Aulsan): Johnston, Trevor, Auslan Corpus Annotation Guidelines, (last update: 2014).